Department of Statistics, University of California-Los Angeles, CA 90095, USA.
Bioinformatics. 2013 Sep 1;29(17):2162-8. doi: 10.1093/bioinformatics/btt365. Epub 2013 Jun 21.
With the accumulation of genome-wide binding data for many transcription factors (TFs) in the same cell type or cellular condition, it is of great current interest to systematically infer the complex regulatory logic among multiple TFs. In particular, ChIP-Seq data have been generated for 14 core TFs critical to the maintenance and reprogramming of mouse embryonic stem cells (ESCs). This provides a great opportunity to study the regulatory collaboration and interaction among these TFs and with other unknown co-regulators.
In combination with liquid association among gene expression profiles, we develop a computational method to predict context-dependent (CD) co-egulators of these core TFs in ESCs from pairwise binding datasets. That is, co-occupancy between a core TF and a predicted co-regulator depends on the presence or absence of binding sites of another core TF, which is regarded as a binding context. Unbiased external validation confirms that the predicted CD binding of a co-regulator is reliable. Our results reveal a detailed CD co-regulation network among the 14 core TFs and provide many other potential co-regulators showing strong agreement with the literature.
See www.stat.ucla.edu/~zhou/CMF for software and source code.
随着越来越多的转录因子（TFs）在同一细胞类型或细胞状态下的全基因组结合数据的积累，系统推断多个 TF 之间复杂的调控逻辑具有重要的现实意义。特别是，已经为维持和重编程小鼠胚胎干细胞（ESCs）所必需的 14 个核心 TF 生成了 ChIP-Seq 数据。这为研究这些 TF 之间以及与其他未知共调控因子之间的调控协作和相互作用提供了很好的机会。
我们结合基因表达谱之间的液体关联，开发了一种计算方法，用于从成对结合数据集中预测 ESCs 中这些核心 TF 的上下文相关（CD）共调控因子。也就是说，核心 TF 和预测的共调控因子之间的共占据取决于另一个核心 TF 的结合位点的存在与否，这被视为结合上下文。无偏外部验证证实了共调控因子的预测 CD 结合是可靠的。我们的结果揭示了 14 个核心 TF 之间详细的 CD 共调控网络，并提供了许多其他与文献强烈一致的潜在共调控因子。
有关软件和源代码，请参见 www.stat.ucla.edu/~zhou/CMF。
